Discussions about things that are not “open source” are often resolved by postponing open source initiatives (OSIs). The software is then open source.
But when you fall into the legal definition of nuts and bolts, and when you fall into the “spirit” that open source actually means, the water is muddy. Certainly there is a significant nuance in the discussion of open source and proprietary software. Are “open source companies” haming projects by sliding the core features behind the commercial paywall? How transparent is there about project development? And how much direct input does “community” actually have in a particular project?
For many, open source is not just about the legal ability to use and modify code. Culture, transparency, and governance around it are of paramount importance.
Everyone knows about Google-flavored versions of Android that ship to smartphones and tablets. The underlying Android Open Source Project (AOSP) will be released under the acceptable Apache 2.0-License, allowing anyone to access the “fork” and change it to their own hardware project.
Android is open source, by almost every definition, to the point that you get. And Google uses this fact to defend against anti-competitive criticism, noting that Amazon has reused Android for its own lineup of fire brand devices. But all this ignores separate “fluoridation agreements” signed with hardware manufacturers that restrict the use of Forked versions of Android. And unlike things like Kubernetes, which are under an independent foundation with various corporate and community contributors, Android is direct control of Google without making roadmap or community input significantly transparent Sitting underneath.
“In the sense of license, Android is perhaps the best-documented, fully open ‘thing’ thing’,” said Luis Villa, co-founder and general counsel of Tidelift, to open CON25’s panel discussion at Open Con25. I mentioned it in. London this week. “All licenses are exactly the same as you want, but I wish you good luck, and when there’s even a next release, I wish you good luck.”
This reaches the heart of the discussion. Open source can be something like an illusion. The lack of true independence means a shortage of agencies for those who want to properly participate in the project. It can also raise questions about the long-term viability of projects, as evidenced by countless open source companies that have switched licenses to protect their commercial interests.
“When you think about the practical accessibility of open source, it’s beyond license, right?” Peter Zaitsev, founder of Percona Percona, said in a panel discussion. “Governance is very important because if you’re a single company, you can change your licenses like ‘it’. ”
These sentiments were echoed in another talk by Dotan Horowitz, an open source evangelist at Cloud Native Computing Foundation (CNCF), thinking about the open source “turn to the dark side.” He noted that in most cases problems arise when a single vendor project decides to make changes based on its own business needs amid other pressures. “Who asks the question, is the open source that is owned by the vendor contradictory?” Holobit said. “I’ve been asking this question for several years, and in 2025 this question is more relevant than ever.”
AI Factor
These debates will not go anywhere anytime soon, as open source emerged as the main focus of the AI realm.
China’s Deepseek has arrived from behind the scenes of open source hype. The MIT license for the model is highly recognized as open source, but among other components there is still a black hole around the training data. This is why Face researchers are trying to create a “more open” version of Deepseek’s inference model.
Meanwhile, Meta has long led open source horns in regards to the Lama brand’s leading language model (LLM), but while Llama is not open source by most estimates, the model is probably “open” than the other models. However, it is probably “open” but commercial restrictions.
“There are whims and concerns about the definition of open source AI, but it’s really clear that what Lama is doing is not open source,” Villa said.
Emily Omier, an open source business consultant and host of open source podcasts, attempts to “damage” the meaning behind “open source” are evidence of its inherent power.
“It shows how strong an open source brand is. The fact that people are trying to corrupt it means people care,” Omier said in a panel discussion. .
However, many of this could be for regulatory reasons. The EU AI Act has special sculptures of “free and open source” AI systems (aside from those considered to pose “unacceptable risks”). And Villa says this is some way to explain why the company wants to rewrite its rulebook as to what “open source” actually means.
“We have a lot of actors now because of brand equity. [of open source] And I would like to change the definition of regulatory meaning. That’s terrible,” Villa said.
Clear parameters
There is a clear argument for applying additional criteria that incorporate the “spirit” of what open source is intended, but having clear parameters defined by the license makes things simple and subtle subjectivity reduces the impact of
How much community engagement does something really need to be “open source”? On a practical and legal level, it makes sense to limit the definition to a license.
OSI executive director Stefano Maffulli said that while some organizations and foundations are leaning towards ideas about “open design, community and development,” these are all fundamentally philosophical concepts.
“The key to having a definition is to have a standard that can be graded, and focusing on licensing is how it is achieved,” Mahuli said in a statement issued to TechCrunch. . “The global community and industry are now relying on the definition of open source. Nowadays, open source AI definitions are an objective measure that can be relied on.”
Source link
